E-tree is an Ethernet service defined by Metro Ethernet Forum (MEF) as a particular rooted multipoint form of an Ethernet Local Area Network (LAN) service. In an e-tree service network, there are two types of network nodes: (1) root nodes; and (2) leaf nodes. A packet sent by a root node may be delivered to other root nodes and other leaf nodes. However, a packet sent by a leaf node is restricted to be delivered only to root nodes in the e-tree service network. In other words, a packet sent by a leaf node is not allowed to be delivered to another leaf node. A packet originating from a leaf node destined for another leaf node should not be delivered in an e-tree service instance.
In conventional VPLS networks, each Provider Edge (network element PE) network element establishes a pseudowire (PW) to every other PE in the network to form a full mesh of PWs between all the PEs in the network. A packet send from any network node may be delivered to any other network node. In this sense, all network nodes in a conventional VPLS network behave like a root node. Hence, in order to implement an e-tree service in a VPLS network, the network element PEs in the network must have a way to differentiate between packets originating from a root node and packets originating from a leaf node, such that any packet originating from a leaf node destined for another leaf node is not forwarded on by the network element PE.
One mechanism to differentiate between packets originating from a root node and packets originating from a leaf node is to modify a control word in each packet to identify the packet as originating from a root node or a leaf node. However, in order to process the modified control word in a network element PE, changes to existing network element PE's internal hardware are required, because the network element PE has to look up an extra field beyond the label stack to perform branching-decisions based on the contents of the control word to determine whether to forward the packet or not.